AI/LLMs AI art bans are going to ruin small 3rd party creators

[dragoner]

I am inclined to agree with you. People hate AI imagery. If they didn't, I might have considered using it for less important pieces. My game-in-progress currently has all work by me and one other person I hired for a few pieces, plus some CC0 textures I modified, and a bit of CC0 floral ornamentation, and I might add in some CC0 nature photography I filtered (the old fashioned way, no AI) to make it look like watercolour. I was planning to buy some 3d models and render them in blender for the places where stock art could be acceptable. You do not need to use AI Image generation, and people do truly hate it.

+1 for blender, they are a good org and should be supported, I have used blender artists for that reason.

 

[Blue]

Yes, and the US Copyright Office agrees.

Okay, if you insist that the US Copyright office agrees that your time has no value, I'll believe you.

But really, what you posted, while true, was orthogonal to the point that was being made.

 

[FrogReaver]

I would hesitate to call it a skill, because you don't have to be anything resembling a polished writer for a prompt to work in an LLM. At best, you can argue that you just need to write enough details of what you're imagining. But the AI is still doing all of the work after that. Unlike the photographer working in reality, who has to have that imagination and actually execute it entirely through human hands.

Do you think someone can imagine a particular image and prompt an ai repeatedly until it produces exactly what they are imagining? (I do). And if thats possible, even for one single person and one single image, then isn’t that the human creative process at work and not the ai?

 

[VHawkwinter]

This "humans and machines aren't the same" argument is how we eventually get killed by our robot overlords.

There is more than that at play here.

There will be a day when we will have have forum arguments about whether or not some human created thing is alive.

I don't think there is something fundamental to humanity that means we will always and forever be the only spark of creativity.

Certainly not. Chimps are apparently in the stone age now. Orcas and Octopodes have societies of a sort. Arguably also crows. Though society doesn't recognise it we already aren't the only people on the planet, just the most technologically advanced of the bunch.

Of course I'm speaking more theoretically and we aren't there YET...but we will be having that conversation one day...or our children's children will.

So, I have a background in software development and more than a minor in psychology. I don't think the fact that our LLM tech isn't a real AGI is primarily a tech limitation. IMO it's a deliberate design decision to NOT make it an AGI. I think we have the tech already to build a proper AGI (and have a pretty reasonable outline of what the components needed would look like), and I suspect some billionaire already has one running on a server rack in their basement. But no LLM like ChatGPT has the correct components (on purpose), and while a real AGI would include an LLM, that would only be one of many components of an AGI.

In my personal opinion Andy Warhols soup cans already have shown that stolen and reworked visuals can be considered proper art.

Collage art. Sure. But that's all made with human decisions, and human decisions are what makes something legally copyrightable.

 

[Morrus]

Do you think someone can imagine a particular image and prompt an ai repeatedly until it produces exactly what they are imagining? (I do). And if thats possible, even for one single person and one single image, then isn’t that the human creative process at work and not the ai?

Describing what you want and making what you want are not the same thing. I think it’s pure arrogance to describe something to a machine which then plagiarises artists to display the thing you described and then claim you created it. It’s like ordering something at a restaurant and then claiming you made the meal because you asked for it.

There’s lots of things I can describe. That doesn’t mean I am making them.

Sadly, making a million dollars is a lot harder than a describing a million dollars.

 

[Gladius Legis]

Do you think someone can imagine a particular image and prompt an ai repeatedly until it produces exactly what they are imagining? (I do). And if thats possible, even for one single person and one single image, then isn’t that the human creative process at work and not the ai?

No, because you are not the one creating that image. You're just telling a third-party creator, in this case an AI, what to do. It's no different than standing over a human artist's shoulder nitpicking at what that artist is painting until it's to your specifications. You did not paint that.

 

[VHawkwinter]

Do you think someone can imagine a particular image and prompt an ai repeatedly until it produces exactly what they are imagining? (I do). And if thats possible, even for one single person and one single image, then isn’t that the human creative process at work and not the ai?

There are people using AI workflows more sophisticatedly that that, combining it with sketches and posed models and Poseable stick figures and photos and character reference sheets and then a prompts, and then refining the image that was made. Those people are approaching an actual art workflow. Look up "ComfyUI" on YouTube. (I bet that's the sort of thing Hasbro is using these days). It's not 2021 when people are just typing prompts into a box and hoping for the best. NVidia's new DLSS 5 is an AI filter running on each frame before it renders as a post-processing layer on actual game renders (but many people are complaining at how it changes character designs, I have saved some fun memes mocking it). There's a Blender plugin (don't remember the name) that runs an AI prompt as an image processing layer on your blender scene renders. The tech now exists in such a way that I think it could be used as an actual art tool, by artists in combination with other techniques especially. (I keep up with tech news and check in on things as they develop, I do not have a ComfyUI setup installed, though I did briefly try it on my 3d Modeling PC.)

But people REALLY hate AI images. Even if you can make actual AI art with human intent now, it's still basically radioactive. I would not recommend it for anything you hope to sell.

 

[jolt]

Just as a point of interest, the US Supreme Court has repeatedly refused to hear challenges regarding AI art and copyright. This has allowed lower court rulings to stand and has effectively reinforced the US Copyright Office's stance that human authorship is required for copyright protection.

Regarding the title of this thread, it's just not true. Third party creators have been very prolific long before the issue of AI generated art has come up. The result of this will be the same as it's always been - such projects will have little art.

 

[VHawkwinter]

But people REALLY hate AI images. Even if you can make actual AI art with human intent now, it's still basically radioactive. I would not recommend it for anything you hope to sell.

As an example of it being used for actual art:

Undergrads the movie. The owner is just the creator now, no MTV. He did not have the funding to do things the old way with 2d animators, he crowdfunded the film. They trained an AI to take blender scenes and re-render them in the style of HD versions of the original animation style, with lots of character models etc. They couldn't use the original show frames for training data for technical reasons so they had to make custom training data for the AI.

Anyway. So they're using the AI to post-process 3d renders as the style of 2d animation they drew in the series 25 years ago.

I will likely watch it when it comes out. But man, is it ever divisive even using it in that way. Dude's getting a lot of hate.

(So for Realm of Stars I don't use LLMs for any of the writing or use diffusion tech for my art. But I do use LLM tools to help debug code (it finds bugs (which are often like a subtle typo) much faster than I can) and sometimes for preliminary code templating which I then rewrite and rework, and to point out editing and grammatical errors in my text. An LLM or similar can also be helpful when you need to look something up but don't know what it's called. Useful for rote tasks that take time, but not for creative work. I'm going to use pattern matching and searching tools, because programmer (I don't personally know any programmers who avoid LLM-based productivity tools). But I don't want a machine doing the creative work, and I think: neither do players / customers).

 

[FrogReaver]

No, because you are not the one creating that image. You're just telling a third-party creator, in this case an AI, what to do. It's no different than standing over a human artist's shoulder nitpicking at what that artist is painting until it's to your specifications. You did not paint that.

Define ‘creating’? Because to me what I described is the act of creation and ai as a tool.